def on_parse_node(self):
super().on_parse_node()
self.name_node = get_tag_el_text(self.oozie_node, "name-node")
self.params_dict = extract_param_values_from_action_node(
self.oozie_node)
_, self.hdfs_files = self.file_extractor.parse_node()
_, self.hdfs_archives = self.archive_extractor.parse_node()
def test_extract_param_values_from_action_node_should_support_el_value(
self):
props = PropertySet(config={},
job_properties={
"userName": "******",
"examplesRoot": "TEST_EXAMPLE_ROOT"
})
# language=XML
xml_content = """
<fragment>
<param>INPUT=/user/${userName}/${examplesRoot}/apps/hive/input/</param>
<param>OUTPUT=/user/${userName}/${examplesRoot}/apps/hive/output/</param>
</fragment>
"""
node = ET.fromstring(xml_content)
result = extract_param_values_from_action_node(node, props=props)
self.assertEqual(
{
"INPUT":
"/user/TEST_USERNAME/TEST_EXAMPLE_ROOT/apps/hive/input/",
"OUTPUT":
"/user/TEST_USERNAME/TEST_EXAMPLE_ROOT/apps/hive/output/",
},
result,
)
def _parse_oozie_node(self):
self.resource_manager = get_tag_el_text(self.oozie_node, TAG_RESOURCE)
self.name_node = get_tag_el_text(self.oozie_node, TAG_NAME)
self.script_file_name = get_tag_el_text(self.oozie_node, TAG_SCRIPT)
self.params_dict = extract_param_values_from_action_node(
self.oozie_node)
self.files, self.hdfs_files = self.file_extractor.parse_node()
self.archives, self.hdfs_archives = self.archive_extractor.parse_node()
def test_extract_param_values_from_action_node_should_return_empty_dict_when_not_found(
self):
props = PropertySet(config={},
job_properties={
"userName": "******",
"examplesRoot": "TEST_EXAMPLE_ROOT"
})
node = ET.fromstring("<dummy></dummy>")
result = extract_param_values_from_action_node(node, props=props)
self.assertEqual({}, result)
def test_extract_param_values_from_action_node_should_parse_single_value(
self):
# language=XML
xml_content = """
<fragment>
<param>INPUT=VALUE</param>
</fragment>
"""
node = ET.fromstring(xml_content)
result = extract_param_values_from_action_node(node)
self.assertEqual({"INPUT": "VALUE"}, result)
def test_extract_param_values_from_action_node_should_parse_single_value(
self):
props = PropertySet(config={}, job_properties={})
# language=XML
xml_content = """
<fragment>
<param>INPUT=VALUE</param>
</fragment>
"""
node = ET.fromstring(xml_content)
result = extract_param_values_from_action_node(node, props=props)
self.assertEqual({"INPUT": "VALUE"}, result)
def _parse_oozie_node(self):
res_man_text = self.oozie_node.find("resource-manager").text
name_node_text = self.oozie_node.find("name-node").text
script = self.oozie_node.find("script").text
self.resource_manager = el_utils.replace_el_with_var(res_man_text,
props=self.props,
quote=False)
self.name_node = el_utils.replace_el_with_var(name_node_text,
props=self.props,
quote=False)
self.script_file_name = el_utils.replace_el_with_var(script,
props=self.props,
quote=False)
self.params_dict = extract_param_values_from_action_node(
self.oozie_node, props=self.props)
self.files, self.hdfs_files = self.file_extractor.parse_node()
self.archives, self.hdfs_archives = self.archive_extractor.parse_node()
def on_parse_node(self):
super().on_parse_node()
self._parse_config()
self.query = get_tag_el_text(self.oozie_node, TAG_QUERY)
self.script = get_tag_el_text(self.oozie_node, TAG_SCRIPT)
if not self.query and not self.script:
raise ParseException(
f"Action Configuration does not include {TAG_SCRIPT} or {TAG_QUERY} element"
)
if self.query and self.script:
raise ParseException(
f"Action Configuration include {TAG_SCRIPT} and {TAG_QUERY} element. "
f"Only one can be set at the same time.")
self.variables = extract_param_values_from_action_node(self.oozie_node)
_, self.hdfs_files = self.file_extractor.parse_node()
_, self.hdfs_archives = self.archive_extractor.parse_node()
def test_extract_param_values_from_action_node_should_support_el_value(
self):
# language=XML
xml_content = """
<fragment>
<param>INPUT=/user/${userName}/${examplesRoot}/apps/hive/input/</param>
<param>OUTPUT=/user/${userName}/${examplesRoot}/apps/hive/output/</param>
</fragment>
"""
node = ET.fromstring(xml_content)
result = extract_param_values_from_action_node(node)
self.assertEqual(
{
"INPUT":
"/user/{{userName}}/{{examplesRoot}}/apps/hive/input/",
"OUTPUT":
"/user/{{userName}}/{{examplesRoot}}/apps/hive/output/",
},
result,
)
def test_extract_param_values_from_action_node_should_return_empty_dict_when_not_found(
self):
node = ET.fromstring("<dummy></dummy>")
result = extract_param_values_from_action_node(node)
self.assertEqual({}, result)